DataSprint 2026 Predicting financial outcomes of Kenyan adults using the 2024 FinAccess Household Survey. SDC × iLab Africa.
# 🇰🇪 Predicting Financial Outcomes in Kenya
### DataSprint 2026 — Strathmore Data Community × iLab Africa
> *One Week. Real Data. Real Impact.*
**Author:** Roy Kimani (@Kimaniroy0)
**Competition:** SDC DataSprint 2026
**Dataset:** 2024 FinAccess Household Survey — 20,871 Kenyan adults across all 47 counties
**Task:** Multiclass classification — predict whether a person's financial situation has *Improved*, *Stayed the same*, or *Worsened*
---
## 📌 Table of Contents
1. Background & Problem Statement
2. Repository Structure
3. Dataset
4. Tools & Libraries
5. Methodology
6. Key EDA Findings
7. Modelling
8. Results
9. Key Drivers of Financial Status
10. Recommendations
11. Challenges & How I Addressed Them
12. How to Run
---
## Background & Problem Statement
Kenya has made significant progress in financial inclusion over the past decade — mobile money has transformed how millions save, borrow, and transact. But the 2024 FinAccess Household Survey tells a harder story: **9.9% of Kenyan adults remain fully excluded from financial services**, and more strikingly, **52.6% of those surveyed said their financial situation had worsened** compared to the previous year.
This project uses machine learning to answer a simple but urgent question:
> **Which factors most strongly predict financial deterioration among Kenyan adults — and what should policymakers, banks, and NGOs do about it?**
The model predicts one of three outcomes for each individual:
| Class | Meaning | Share of Dataset |
|-------|---------|-----------------|
| **Worsened** | Financial situation deteriorated vs. last year | 52.6% |
| **Stayed the same** | No significant change | 26.9% |
| **Improved** | Financial situation improved vs. last year | 20.5% |
---
## Repository Structure
```
kenya-financial-status-datasprint2026/
│
├── notebooks/
│ └── DataSprint2026_Notebook.ipynb # Full pipeline: EDA → Modelling → Interpretation
│
├── visualisations/
│ ├── chart1_target_distribution.png # …